基于改进多目标灰狼算法的装配线平衡与预防维护集成优化

Integrated optimization of assembly line balance and preventive maintenance based on improved multi-objective grey wolf algorithm

在传统的装配线运行管理中,装配线平衡和设备预防维护是完全分离的两部分,造成了生产的巨大损失和维护的极大不便。面向正常工作的节拍最小化和给定预防维护情形下的节拍最小化及生产调整最小化3个目标,构建装配线平衡与工位预防维护的集成优化模型,以期制定出能提供设备维护契机、保证维护时生产连续性的多套分配方案。提出了面向多目标优化的改进灰狼算法,其中基于集成问题的离散特征,设计了基于随机键的编码和解码方法;融合Pareto层级构造和拥挤距离,提出了狼群等级划分方法和更新机制,以保留并更好地探索近优解;引入了交叉算子,以提高算法在每个案例中的分布性。实验结果表明改进后灰狼算法的收敛性、分布性较好,所获得非支配解集更逼近Pareto最优前沿,且平衡与维护的有效集成可显著降低生产成本。

In the traditional assembly line operation management,the assembly line balance and the equipment maintenance on the station are completely separated,which resulting in huge production losses and great inconvenience in maintenance.Aiming at minimizing the cycle time and production adjustment in normal work and given preventive maintenance situations,an integrated optimization model of assembly line balancing and predictive maintenance was constructed to generate multiple allocation plans that could provide maintenance opportunities and promote the production continuity,and an improved grey wolf algorithm was proposed for this multi-objective optimization problem.Considering discrete features of the integration problem,a random key-based encoding and decoding method was employed.Facing with the multi-objective optimization characteristics,a wolf class classification method and update mechanism were designed,which integrated Pareto hierarchy and congestion distance,so as to preserve and better explore the near-optimal solution.A crossover operator was hired to improve the diversification of grey wolf optimization algorithm.Experimental results showed that the improved grey wolf algorithm had better convergence and diversity,and the obtained non-dominated solutions approach was closer to the Pareto optimal frontier.The effective integration of balancing and maintenance could significantly reduce the production costs.